A lightweight, concurrent Kubernetes operator built in Rust using the kube-genops and kube-objstore crates. It automates disaster recovery (DR) storage states by syncing PersistentVolume (PV) metadata across protected and recovery clusters via object storage.
The relink-operator solves the "orphaned volume" problem during cross-cluster failovers. Instead of manually patching bi-directional storage bindings (spec.claimRef) during an outage, this operator splits the lifecycle into two concurrent, decoupled controller loops managed by Custom Resource Definitions (CRDs):
VolumeRelinkSource(Protected Cluster): Watches active PVC/PV states matching specified selectors, captures their volume mapping metadata, and pushes state snapshots to an S3-compatible object storage bucket.VolumeRelinkTarget(Recovery Cluster): Sits in a standby loop monitoring the object storage bucket. When a failover event is explicitly triggered, it automatically pulls down the metadata, clears out conflicting binding definitions, and pre-binds the existing cloud volumes to newly instantiated PVCs.